F-values.txt

F-value is the determinant for chasing. It is a multi-variant function of Number of Remaining Students(NR), Students' properties (including career and position) and the Tricker's own properties(CD, position...).
Let the position of the target Student be Xs, and that of the Tricker's be Xt.
We should take the approximity of the Student and Machines, Tricker and other Students into account.
As different careers affect behavior in completely different directions, it should be a parameter in a switch-case procedure.

If K == 'Teacher':
	F = (NR==1)? (\frac{ |Xt - Xm| + a0}{a1 * |Xt - Xs|} : 0;
	return;
	where a0, a1 are real parameters and Xm is the position of the nearest Machine;
	//value 0 is to completely avoid the possibility of fixation on the Teacher

Else:
	//No need to consider the Student's skills: that's the work of other functions
	F(NR, Xs, Xt, Xm, parameter_array(career))
	R = 2 * log0.5(NR - ND) + 5 might be an appropriate coefficient with regards to NR, where ND is number of Deceased Students.
	other factors:
	RD = \frac{a2(career)}{|Xt - Xs|} is relevant to the relative distance between Tricker and Target;
	Cluster = sum(\frac{a3(career, Stu1) * a3(career, Stu2)}{|Xs1 - Xs2| + min_distance}) is the cluster bonus. As this is unimportant, it is only additional.
	Blood = (Within sight) ? Blood_value : Full_blood - (Full_blood - last_Blood_value)*0.998^(frame_since_last),
	considering the possibility of being replenished
	Dir = 1 + a1(career) * cos<Xt-Xm, Xt-Xs> / |Xt-Xm| //Reuse a1

	Thus propose F = R * RD * (1 - a4(career) * Blood) * Dir + Cluster, which can be extended to
	F = (2 * log0.5(NR - ND) + 5) * frac{a2(career)}{|Xt - Xm|} * (1 - a4(career) * ((Within sight) ? Blood_value: (Full_blood + (Full_blood - last_Blood_value)*0.998^(frame_since_last)) )) * 1 + a3 * \frac{cos{<Xt-Xm, Xt-Xs>}}{|Xt-Xm|} + sum(\frac{a3(career, Stu1) * a3(career, Stu2)}{|Xs1 - Xs2| + min_distance})